I suspect but can't prove that these markings are not adaptive, but are accidents or epiphenomena of the general genetic system that determines coat colour.
On the one hand, an Australian Broadcasting Corporation documentary claims that canid eyebrows are adapted for social interaction: they have specialized muscles (which are not present in other carnivorous mammals such as cats), and ...
One study showed foxes who hunt alone had about half the facial expressions of wolves who work in packs. In fact, in wolves and dingoes, the eyebrows are often even a different colour, exaggerating the movement.
(they don't give any primary references).
On the other hand (supporting my guess) a lot of the evolution of these colour patterns seems to have occurred after domestication. These markings (not the eyebrow spots in particular) seem to be called tan points:
Red (tan) appears as pips above the eyes, on the sides of the muzzle extending to the cheeks, as pips on the cheeks, on the front of the neck just below the head, as two triangular patches on the front of the chest, on the lower legs and feet (and inside of the legs), and as a patch underneath the tail (and sometimes along the bottom edge of the tail too).
(I don't know whether eyebrow contrast in animals that aren't strictly brown/black is caused by the same genes ...)
A PhD thesis by Dayna Dreger (some material published in Journal of Heredity 2013:104(3):399–406 doi:10.1093/jhered/est012)
The black-and-tan phenotype, associated with the at allele, is a predominantly eumelanistic phenotype, with phaeomelanin restricted to distinct regions on the lower limbs, cheeks, eyebrows, chest, and around the anus. (p. 7)
Genome wide association study analysis, fine mapping, and sequence analysis identified a 16 bp tandem duplication in intron 5 of the hnRNP-Associated with Lethal Yellow (RALY) gene that segregates with the black-and-tan phenotype, versus saddle tan, in Basset Hounds and Pembroke Welsh Corgis. In breeds that never have the saddle tan phenotype, but frequently have the black-and-tan phenotype, the RALY duplication does not segregate with black-and-tan. This, together with further genotype analysis, suggests a gene interaction of ASIP, MC1R, DEFB103, RALY, and an additional modifier gene is required for expression of saddle tan or black-and-tan. (p. ii)
Neither the coyote nor 9 of the 10 wolves had the RALY duplication and the tenth atypical wolf was heterozygous (+/dup) (Table 5.1), suggesting that the lack of the duplication is the ancestral or wild-type allele (+). It follows then that the saddle tan phenotype is ancestral to the black-and-tan phenotype, despite the relatively limited number of breeds that currently express the saddle tan phenotype, making black-and-tan a modification of the saddle tan phenotype. The popularity of the black-and-tan phenotype across breeds and breed types is likely explained by artificial selection for the striking black-and-tan phenotype over that of saddle tan. Since the saddle tan pattern is found primarily in terriers, scent hounds, and a small number of herding breeds, this finding may also shed light on the development and relation of modern dog breeds, hinting at common ancestors between breed types.
